342 patients referred to a neurological out-patient department in North East England between November 1975 and November 1976 were studied prospectively using two methods of psychiatric case identification, by clinical interview and by use of the General Health Questionnaire. The population of psychiatric patients identified by clinical techniques is different from that identified by the GHQ and it is concluded that GHQ cannot be used for routine screening in the neurological clinic. Clinical data were compared with a previous retrospective study. Clinical assessment revealed 27% with psychiatric disorders whilst GHQ showed a probable psychiatric morbidity of 48% with a cut off score of 11/12 and 21% with our best cut off score of 26/27. Statistical problems with this method are discussed and the results of this study compared with others using the same questionnaire. Aspects of organic and psychiatric disorder in a neurological clinic are discussed.